.. it seems to me that, properly understood, a "policy choice" in a Dworkinian universe only means cheating your own considerate decision constituence. If the "right answer" in a case is how your coherent views about law come to compel the best choice that is a function of those views, then "violating law" simply means violating your own considerate decision constituence (being internally contradictory) -- or, perhaps, not having a "constituence" to begin with.
Properly understood, Dworkin is a structuralist. And a Kantian one at that. For one to "cheat law," one would have to either cheat one's own considerate views, not have such views, not be genuine when declaring the conclusion of those views -- or, perhaps, just have a bad day at work.
I think it is a mistake to read Dworkin as saying that Hercules represents anything more than (a) considerate views; (b) coherence and structure in the actions upon them; (c) necessarily compelled answers as a result of (a) plus (b). Hercules does NOT represent one true answer in a case for all possible judges. That's a sloganized reading. He represents his OWN best CONCLUSION derived from his legitimate system of thought about what "law" requires. Here is a good passage that compares quite nicely to Wittgenstein's views on aesthetics:
"We can usefully compare the judge deciding what the law is on some issue not only with citizens of courtesy deciding what that tradition requires, but with the literary critic teasing out the various dimensions of value in a complex play or poem." p. 228, LE.
So we have to ask: What is Posner talking about? This is a snag in the language game. "Voting policy" or "voting ideology" where those terms mean "violating law" -- where that means "not judging correctly" -- only means violating your own considerate system of thought, or not having a considerate system. I doubt Posner says this is what he does.